Page 1
RESEARCH ARTICLE
Effect of ice slushy ingestion and cold water
immersion on thermoregulatory behavior
Hui C. ChooID1☯*, Jeremiah J. Peiffer2☯, João P. Lopes-Silva3☯, Ricardo N.
O. MesquitaID1‡, Tatsuro Amano4‡, Narihiko Kondo5‡, Chris R. Abbiss1☯
1 Centre for Exercise and Sports Science Research, School of Medical and Health Sciences, Edith Cowan
University, Joondalup, Western Australia, Australia, 2 School of Psychology and Exercise Science, Murdoch
University, Murdoch, Western, Australia, Australia, 3 School of Physical Education and Sport, University of
São Paulo (USP), São Paulo, São Paulo, Brazil, 4 Faculty of Education, Niigata University, Niigata, Niigata
Prefecture, Japan, 5 Laboratory for Applied Human Physiology, Graduate School of Human Development
and Environment, Kobe University, Kobe, Hyōgo Prefecture, Japan
☯ These authors contributed equally to this work.
‡ These authors also contributed equally to this work.
* [email protected]
Abstract
Two studies were conducted to examine the effects of ice slushy ingestion (ICE) and cold
water immersion (CWI) on thermoregulatory and sweat responses during constant (study 1)
and self-paced (study 2) exercise. In study 1, 11 men cycled at 40–50% of peak aerobic
power for 60 min (33.2 ± 0.3˚C, 45.9 ± 0.5% relative humidity, RH). In study 2, 11 men cycled
for 60 min at perceived exertion (RPE) equivalent to 15 (33.9 ± 0.2˚C and 42.5 ± 3.9%RH).
In both studies, each trial was preceded by 30 min of CWI (~22˚C), ICE or no cooling (CON).
Rectal temperature (Tre), skin temperature (Tsk), thermal sensation, and sweat responses
were measured. In study 1, ICE decreased Tre-Tsk gradient versus CON (p = 0.005) during
first 5 min of exercise, while CWI increased Tre-Tsk gradient versus CON and ICE for up to
20 min during the exercise (p<0.05). In study 2, thermal sensation was lower in CWI versus
CON and ICE for up to 35–40 min during the exercise (p<0.05). ICE reduced thermal sensa-
tion versus CON during the first 20 min of exercise (p<0.05). In study 2, CWI improved
mean power output (MPO) by ~8 W, compared with CON only (p = 0.024). In both studies,
CWI (p<0.001) and ICE (p = 0.019) delayed sweating by 1–5 min but did not change the
body temperature sweating threshold, compared with CON (both p>0.05). Increased Tre-Tsk
gradient by CWI improved MPO while ICE reduced Tre but did not confer any ergogenic
effect. Both precooling treatments attenuated the thermal efferent signals until a specific
body temperature threshold was reached.
Introduction
During self-paced exercise with increased exogenous heat load, behavioral thermoregulation
can be achieved through adjusting the work rate to manipulate metabolic heat production
since heat dissipation is limited by involuntary mechanisms, i.e., sweating and cutaneous
PLOS ONE | https://doi.org/10.1371/journal.pone.0212966 February 27, 2019 1 / 19
a1111111111
a1111111111
a1111111111
a1111111111
a1111111111
OPEN ACCESS
Citation: Choo HC, Peiffer JJ, Lopes-Silva JP,
Mesquita RNO, Amano T, Kondo N, et al. (2019)
Effect of ice slushy ingestion and cold water
immersion on thermoregulatory behavior. PLoS
ONE 14(2): e0212966. https://doi.org/10.1371/
journal.pone.0212966
Editor: Caroline Sunderland, Nottingham Trent
University, UNITED KINGDOM
Received: October 28, 2018
Accepted: February 12, 2019
Published: February 27, 2019
Copyright: © 2019 Choo et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
author and source are credited.
Data Availability Statement: All relevant data are
within the manuscript and its Supporting
Information files.
Funding: The study was funded by the School of
Medical and Health Sciences, Edith Cowan
University. At the time of the study was conducted,
HCC and RNOM were supported by the
International Postgraduate Research Scholarship
and Edith Cowan University. João Lopes-Silva
(88881.132395/2016-01) was supported by
Page 2
vasodilation [1]. Under this paradigm, skin and core temperature and thermal perception have
been identified as controllers of thermoregulatory behavior [2–4]. These controllers may mod-
ify thermoregulatory behavior independently [3, 5]. Indeed, improved thermal sensation and
greater work output have been observed following oral L-menthol mouth rinse [5], face cool-
ing via fan or application of topical menthol gel [3], without changes in the skin and core
temperature.
External cooling via whole body cold water immersion (CWI) and internal cooling via
ingestion of ice slushy (ICE) have often been used as ergogenic aids before exercise in the heat
[6]. The aim of precooling is to create a greater heat sink for subsequent metabolic heat pro-
duction; however, the ergogenic mechanisms underlying precooling may be specific to differ-
ent cooling methods. For example, whole body CWI involves direct contact with a large body
surface area and thus resulting in reduced skin blood flow and increased core-to-skin tempera-
ture gradient, whereas ICE has a more direct effect on the core body temperature [7]. Recent
meta-analyses have shown that CWI improves aerobic exercise performance in warm environ-
ments [6], with a lesser beneficial effect observed in moderate conditions (i.e., ambient temper-
atures between 18–26˚C) [8]. During exercise in temperate environments, drastic muscle
cooling during CWI has been shown to negatively impact delivery of oxygen (O2) and sub-
strates to locomotive muscle as assessed by near-infrared spectroscopy (NIRS), resulting in
increased anaerobic metabolism during subsequent exercise in temperate environments [9,
10]. It is worth noting that exercise in the heat has also been shown to impair muscle blood vol-
ume and tissue oxygenation assessed by NIRS [11].
ICE is logistically less challenging than CWI and has minimal muscle cooling effect, making
this cooling method preferable to CWI. However, a recent meta-analysis showed that CWI
and ICE may elicit different influences on the thermoregulatory behavior during exercise in
the heat via changes in skin and core temperatures and thermal perception [6]. Specifically,
while both ICE and CWI effectively reduced core temperature, but only CWI had a clear bene-
ficial effect on exercise performance concomitant with reductions in skin temperature and
thermal sensation. Additional benefits of CWI versus ICE include a greater body surface area
being directly cooled during whole body immersion, a continued cooling effect after immer-
sion [12], the practicality of ICE being limited by an optimal drinking volume required for sig-
nificant core body cooling effect [13], and improved whole body fluid balance through lesser
sweat loss [6]. Sweating during exercise is initiated by both thermal and non-thermal factors,
with nitric oxide being identified as a non-neural regulating factor [14, 15]. ICE has been
shown to activate the intra-abdominal thermoreceptors but minimally affect skin temperature
and skin blood flow [16, 17]. Hence, sweating during ICE is predominantly regulated by ther-
mal reflexes, whereas direct skin cooling during CWI is likely to regulate the sweat response
via activation of the thermoreceptors and the nitric oxide pathway through blood flow
reduction.
The present study aimed to examine the differences in some thermoregulatory parameters
(i.e., core-to-skin temperature gradient, sweat response and thermal sensation) and muscle
perfusion (NIRS parameters) during CWI and ICE versus a no cooling control condition
(CON), and the consequential influence on thermoregulatory behavior as indicated by total
work output during exercise in the heat. Accordingly, two studies were conducted. In the first
study, the thermoregulatory parameters and muscle perfusion were assessed during 60 min of
exercise at a fixed intensity following CWI, ICE and CON. In the second study, thermoregula-
tory behavior was investigated during 60 min of cycling at a fixed rating of perceived exertion
(RPE). The RPE clamp protocol allows individuals to modify power output continually based
on the regulatory role of perceived exertion in behavioral thermoregulation [18], and has been
utilised to investigate the differential influences of thermal perception and temperature on
ICE and CWI on thermoregulatory behavior
PLOS ONE | https://doi.org/10.1371/journal.pone.0212966 February 27, 2019 2 / 19
CAPES Scholarship. The other authors have no
support or funding to report.
Competing interests: The authors have declared
that no competing interests exist.
Page 3
thermoregulatory behavior [3, 5]. In the first study, it was hypothesized that CWI would result
in a greater core-to-skin temperature gradient, but would attenuate the sweat responses and
muscle perfusion when compared with ICE and CON. In the second study, we hypothesized
CWI would improve mean power output and total work output relative to CON and ICE.
Materials and methods
Participants
The experimental procedures were approved by Edith Cowan University ethics committee for
human research and were conducted according to the principles expressed in the Declaration
of Helsinki. All participants were recruited locally via the institutional intranet portal and
social media. All procedures and associated risks were made known to the participants before
obtaining signed consent. For study 1, a priori analysis using whole body sweat loss data from
a previous study [19] was performed using the calculated effect size of 0.78, an α of 0.05, and a
β of 0.2. Minimum of 7 participants were required to identify significant difference in whole
body sweat loss between conditions. Nevertheless, 13 men were recruited for study 1; however,
only 11 men completed all trials (mean ± SD; age: 27 ± 6 y, body mass: 77.5 ± 10.5 kg, height:
177.1 ± 7.9 cm, sum of 4 skinfolds: 58.1 ± 25.7 mm, Peak O2 uptake ( _VO2peak): 43.9 ± 11.4
mL�kg-1�min-1, peak aerobic power: 288 ± 64 W). One participant withdrew from the study
due to personal reasons and another could not complete the exercise task. For study 2, power
analysis was determined using an effect size of 0.80 based the work output data from a previous
study using similar RPE clamp protocol [3]. A minimum of 7 participants were required to
identify significant difference between conditions with an α of 0.05 and a β of 0.2. Eleven out
of 13 participants completed the study (mean ± SD; age: 30 ± 6 y, body mass: 80.6 ± 12.8 kg,
height: 1.8 ± 0.1 m, _VO2peak: 51.1 ± 8.2 mL.kg-1.min-1, sum of 7 skinfold: 131.2 ± 52.6 mm).
Two participants did not complete the trials due to personal reasons. Thirteen participants
were recruited for each study after taking into consideration attrition, missing data and inher-
ent differences in the study designs. All participants were recreationally active who engaged in
physical activity (e.g., cycling, running and soccer) for�3 times per week within the past two
years, non-smokers and free from cardiovascular disease. All participants completed a prelimi-
nary visit and three experimental trials. Participants were asked to: 1) avoid strenuous exercise
and alcohol consumption during the 24 h before each trial; 2) avoid caffeine during the 12 h
before each trial; and 3) keep their diet, physical activity and sleep habits consistent before
each trial, assisted by a 1-d dietary intake and physical activity record.
Preliminary measurements
Anthropometric measurements and _VO2peak were assessed during a preliminary session. The
_VO2peak, defined as the highest 30-sec average, was assessed by an incremental cycling exercise
test (Velotron Racermate, Seattle, WA, USA) starting at 70 W and increased by 35 W�min-1
until cadence dropped below 60 rpm. Minute ventilation, carbon dioxide production and _VO2
was measured by a calibrated metabolic cart (TrueOne 2400, ParvoMedics, Utah, USA) during
the incremental exercise test. Peak power output was prorated from the last completed stage
plus the time in the last uncompleted stage multiplied by 35 W [20]. Heart rate ([HR], Polar
S810i, Polar Electro Oy, Kempele, Finland) and RPE [21] were also assessed during the exer-
cise. In study 2, participants performed a standardised familiarisation trial adapted from
Lander et al. [22] after the incremental exercise test. The familiarisation trial began at RPE 11
for 4 min and increased to RPE 13 (3 min), RPE 15 (2 min) and ended at RPE 19 (1 min).
ICE and CWI on thermoregulatory behavior
PLOS ONE | https://doi.org/10.1371/journal.pone.0212966 February 27, 2019 3 / 19
Page 4
Experimental trials
In both studies, participants completed three experimental trials separated by at least 48 h, and
the time of trials was kept within 2 h between sessions for each participant. All participants
arrived 30 min before the experimental trials. Following urine specific gravity index (USG,
Atago hand-held refractometer, Model Master-URC/Nα, Tokyo, Japan), nude body mass mea-
surement (Model ID1, Mettler Toledo, Columbus, Ohio, USA) and self-insertion of a rectal
thermistor, participants proceeded to a regulated climate chamber for at least 20 min for ther-
mal equalisation and further instrumentation
In study 1, participants performed one of the three 30-min pre-exercise treatments in a ran-
domized crossover manner: 1) ingestion of 1.25 g�kg-1�5 min-1 of an ice slushy mixture (-0.7 ±0.1˚C) with added orange flavoured syrup (Cottee’s Foods, NSW, Australia); 2) mid-sternal
level CWI at 22.1 ± 0.1˚C; and 3) passive rest on a chair beside the cycle ergometer during
CON. During CWI and CON, participants consumed 1.25 g�kg-1�5 min-1 of warm fluid (36.3 ±0.6˚C) with added orange flavoured syrup. For all conditions, the mixed drinks contained 6%
carbohydrate. At 9 min 9 sec ± 1 min 31 sec after the end of cooling, participants cycled for 60
min at 40–50% of peak aerobic power at 33.2 ± 0.3˚C, 45.9 ± 0.5%RH. In study 2, participant
completed three trials which involved 30 min of pre-exercise treatments. Before exercise, they
consumed 1.25 g�kg-1�5 min-1 of ICE (0.1 ± 0.1˚C) containing 0% carbohydrate, completed 30
min of CWI at 22.3 ± 0.2˚C, and passive rest (CON) in a randomized, crossover manner. Warm
water (35.8 ± 0.3˚C) was consumed at 1.25 g�kg-1�5 min-1 during CWI and CON. For both stud-
ies, the ice slushy mixture (1:1 mixture of ice and liquid) was prepared using a commercially
available food blender. All drinks consumed during precooling and exercise were stored in an
insulated flask, and temperature of the drinks was measured immediately before serving.
In study 2, the experimental trials required participants to cycle for 60 min at a pace equiva-
lent to 15 or ‘hard or heavy’ on the 15-point RPE scale [21] at 33.9 ± 0.2˚C and 42.5 ± 3.9%RH.
Participants had access to the elapsed time, but no other feedback or verbal encouragement
was given. Exercise commenced at 10 min 22 sec ± 1 min 4 sec after the end of cooling. In
both studies, thermal sensation [23] and RPE [21] were assessed every 5 min, and water con-
sumption during the exercise was matched between trials for each participant.
Temperature and sweat measurements
Temperature and sweat data were logged at 0.2 Hz continuously during the experimental trials
by a Squirrel data logger (Model 2040, Grant Instruments Ltd., Cambridge, UK). Weighted
mean skin temperature was calculated from the measurement of four sites [24] over the ster-
num (Tst), forearm (Tarm), thigh (Tth) and calf (Tca) (YSI 409B thermistors, Dayton, OH,
USA). Rectal temperature (Tre) was measured via a thermistor (Monatherm Thermistor 400
Series, Mallinckrodt Medical, St. Louis, MO, USA) self-inserted 10 cm past the anal sphincter.
Mean body temperature (Tb) was calculated as (0.79 × Tre) + (0.21 × Tsk) [25].
Local sweat rate ([LSR], mg�cm-2�min-1) was measured using ventilated sweat capsules
(5.31 cm2) attached to the left dorsal forearm 5 cm below the antecubital fossa (LSRarm), and
the left thigh 15 cm above the superior border of the patella (LSRth). Dry air ventilated the cap-
sules at 1.5 mL�min-1 and the water content of the effluent air was measured using capacitance
hygrometers (HMP60, Vaisala, Helsinki, Finland). Onset of sweating in terms of exercise time
was determined by fitting the 1-min averaged data using a one-phase exponential association
model [26]. Thermoregulatory sweating threshold and sensitivity were determined by plotting
LSR averaged from both sites against Tre and Tb as previously described [27]. Whole body
sweat loss was determined from the change in body mass to the nearest 10 g, taking into con-
sideration the total volume of fluid consumed.
ICE and CWI on thermoregulatory behavior
PLOS ONE | https://doi.org/10.1371/journal.pone.0212966 February 27, 2019 4 / 19
Page 5
NIRS measurements
A near-infrared spectroscopy system (Niromonitor NIRO-200, Hamamatsu Photonics, Japan)
was used to measure tissue chromophore concentration changes in the right vastus lateralis at
10 Hz (Powerlab 16/30 ML 880/P, ADInstruments, New South Wales, Australia). The NIRS
probe, with an interoptode distance of 4 cm, was secured to the right thigh with the centre of
the probe 15 cm above the superior border of the patella towards the greater trochanter. A
black polyethylene sheet (7 cm × 5 cm × 30 μm) was secured over the NIRS probe to minimise
ambient light contamination, and participants placed their legs in a polyethylene bag (365
cm × 228 cm × 30 μm) to protect the measuring instruments from water damage during CWI.
Using the modified Beer-Lambert method, the NIRS system measures changes (μM�cm-1)
in oxyhaemoglobin (OxyHb), deoxyhaemoglobin (Hb) and total haemoglobin (tHb) at 775,
810 and 850 nm from an arbitrary initial value. Simultaneously, tissue oxygenation index
(TOI, %) is determined based on the spatially resolved spectroscopy method. Changes in the
tHb and TOI reflect muscle blood volume and ratio between oxygenated haemoglobin and
total haemoglobin, respectively [28]. The NIRS data were averaged over 1 min and expressed
as absolute differences from the baseline values.
Heart rate, mean arterial pressure and skin perfusion
HR was recorded continuously during the experimental trials using the same Polar telemetric
monitor. In study 1, skin perfusion was measured at the left forearm by laser Doppler flowme-
try system (PeriFlux 5000 with thermostatic probe 457, Perimed AB, Jarfalla, Stockholm, Swe-
den), expressed as arbitrary perfusion units (PU). However, due to probe displacement during
the exercise, the authors decided not to include the PU signals in the data analysis. In study 2,
skin PU was measured at the left thigh. Distance between the sweat capsule, skin thermistor
and the laser Doppler probe were kept at 2 cm. Mean arterial pressure (MAP) was obtained
from the blood pressure waveform recorded from a finger (Finapres NOVA, Finapres Medical
Systems©, Amsterdam, The Netherlands). The PU and MAP signals were sampled at 10 Hz
using the same data acquisition software as the NIRS signals. MAP was not recorded during
the RPE clamp exercise (study 2) to minimise distraction to the participants. Cutaneous vascu-
lar conductance (CVC) was determined using the quotient between PU and MAP at baseline
and at the end of cooling.
Statistical analysis
Statistical analyses were performed using the nlme and emmeans packages for R v3.5.0 (R
Core Team, R Foundation for Statistical Computing, Vienna, Austria). Linear mixed effects
modeling allowed inclusion of data sets with missing values during the exercise for Tre and the
NIRS signals due to thermistor displacement and probe damage, respectively. The dependent
variables were analyzed in separate linear mixed models fitted with restricted maximum likeli-
hood. The initial models included a by-participant random intercept. Random effects for
intercept and for slope with regards to condition, and covariance between intercepts and
slopes were incorporated where indicated by minimising the values for Akaike information
criteria. Marginal p-values were reported. Bonferroni correction was applied for multiple pair-
wise comparisons at a given time point. Significance level was accepted at p�0.05. A total of 19
data sets from study 1 and 2 were included in the analysis of the sweating threshold and sensi-
tivity. One participant partook in both studies and thus his data set was included only once.
Additionally, two data sets from study 1 were excluded due to water damage during CWI and
due to missing values >30 min during the exercise. All data are expressed as mean ± SD.
ICE and CWI on thermoregulatory behavior
PLOS ONE | https://doi.org/10.1371/journal.pone.0212966 February 27, 2019 5 / 19
Page 6
Results
Study 1
Baseline body mass and USG, physiological and perceptual responses during exercise.
A significant effect was present for baseline body mass (p = 0.006) with post-hoc analysis
revealing that baseline body mass was lower in CON versus ICE (p = 0.008), but no difference
was observed between CWI and CON (p = 0.195) or between CWI and ICE (p = 0.979, CON:
77.3 ± 9.7 kg, CWI: 77.6 ± 10.2 kg, ICE: 77.8 ± 9.9 kg). USG before the exercise was similar
between conditions (p = 0.494, CON: 1.013 ± 0.006 g�ml-1, CWI: 1.010 ± 0.007 g�ml-1, ICE:
1.011 ± 0.008 g�ml-1). Table 1 shows the mean physiological and perceptual responses during
the 60 min of exercise. Whole body sweat loss was significantly lower in CWI compared with
CON (p = 0.014) while there was no difference between ICE and CON (p = 0.161). Mean HR
was lower in CWI relative to CON (p = 0.025), while differences between CWI and ICE did
not reach statistical significance (p = 0.068). Thermal sensation, RPE, LSRarm and LSRth were
not significantly different between conditions (Table 1).
Tsk, Tre and Tre-Tsk gradient. One participant’s data for Tre were removed from the anal-
ysis due to water damage during CWI. A significant interaction effect was observed for Tre
(p = 0.020). ICE decreased Tre by 0.3˚C during the first 5 min of exercise when compared with
CWI and CON (p<0.05, Fig 1A). However, the magnitude of increase in TreΔ during the exer-
cise was higher in ICE relative to CON (p = 0.012) and CWI (p = 0.001, Table 1), resulting in
similar Tre at the end of exercise (Fig 1). An interaction effect was observed for Tsk (p<0.001)
such that it was significantly lower in CWI compared with ICE and CON during the exercise
(Fig 1B). Tre-Tsk gradient exhibited an interaction effect (p<0.001) such that CWI increased
the gradient during the first 15 min of exercise compared with CON and ICE (p<0.05, Fig
Table 1. Mean physiological and perceptual responses during 60 min of cycling at fixed exercise intensity (study 1), and during the RPE clamp exercise (study 2).
CON CWI ICE P-value
Study 1:
Whole body sweat loss (mL) 1244 ± 374� 1064 ± 343 1128 ± 329 0.016
Thermal sensation (AU) 6.0 ± 1.1 5.7 ± 0.6 5.8 ± 0.7 0.384
RPE (AU) 13.5 ± 1.9 13.4 ± 1.8 13.4 ± 1.6 0.982
Heart rate (beats�min-1) 154 ± 14� 149 ± 15 153 ± 14 0.018
TreΔ (˚C) (n = 8) 1.3 ± 0.5# 1.2 ± 0.6# 1.6 ± 0.6 <0.001
LSRarm 1.25 ± 0.75 1.10 ± 0.53 1.14 ± 0.44 0.470
LSRth 0.71 ± 0.46 0.74 ± 0.36 0.69 ± 0.35 0.772
Study 2:
Mean power output (W) 130 ± 20� 138 ± 18 129 ± 25 0.018
Total work output (kJ) 470 ± 74� 498 ± 65 464 ± 90 0.018
Whole body sweat loss (mL) 1394 ± 381� 1239 ± 367 1396 ± 119� 0.004
Heart rate (beats�min-1) 144 ± 20 141 ± 14 144 ± 15 0.679
TreΔ (˚C) 1.4 ± 0.5# 1.2 ± 0.6# 1.7 ± 0.5 <0.001
LSRarm 1.21 ± 0.35 1.21 ± 0.53 1.20 ± 0.60 0.994
LSRth 0.63 ± 0.19 0.68 ± 0.29 0.66 ± 0.27 0.592
CON, control; CWI, cold water immersion, ICE, ice slushy ingestion; RPE, rating of perceived exertion; TreΔ, magnitude of increase in rectal temperature during
exercise; LSRarm, local sweat rate for the arm at the end of exercise; LSRth, local sweat rate for the thigh at the end of exercise
� p<0.05 versus CWI# p<0.05 versus ICE.
Data are mean ± SD for n = 11 unless otherwise stated. See S5 Table for effect sizes (Cohen’s d) calculated from mean differences between conditions and pooled SD.
https://doi.org/10.1371/journal.pone.0212966.t001
ICE and CWI on thermoregulatory behavior
PLOS ONE | https://doi.org/10.1371/journal.pone.0212966 February 27, 2019 6 / 19
Page 7
ICE and CWI on thermoregulatory behavior
PLOS ONE | https://doi.org/10.1371/journal.pone.0212966 February 27, 2019 7 / 19
Page 8
1C). ICE decreased Tre-Tsk gradient relative to CON during the first 5 min of exercise
(p<0.05).
NIRS data. An interaction effect was observed for tHb (p<0.001) whereby CWI signifi-
cantly decreased tHb relative to CON and ICE during the first 20 min of exercise (Fig 2A).
OxyHb exhibited an interaction effect (p<0.001) such that it was lower in CWI versus CON
and ICE during the initial 10 min of exercise (p<0.05, Fig 2B). Hb showed an effect for time
(p<0.001), but no main condition (p>0.05) or interaction effect (p = 0.350, Fig 2C) was
observed. Although TOI demonstrated an effect for time (p<0.001), there was no condition
(p>0.05) or interaction effect (p = 0.580, Fig 2D).
Study 2
Baseline body mass and USG. There was no difference between conditions for baseline
body mass (p = 0.498, CON: 80.2 ± 13.3 kg, CWI: 80.5 ± 12.9 kg, ICE: 80.3 ± 12.9 kg) and USG
(p = 0.255, CON: 1.016 ± 0.009 g�ml-1, CWI: 1.017± 0.007 g�ml-1, ICE: 1.013 ± 0.008 g�ml-1).
Four participants took up to 34 min 32 sec to ingest the given volume in the ICE trials.
MPO, total work output, HR and sweat responses. MPO in CWI was greater than CON
(p = 0.024), but there was no difference between CON and ICE (p>0.999) or between CWI
and ICE (p = 0.263, Table 1). Similarly, total work output in CWI was greater than CON
(p = 0.024), whereas there was no difference between CON and ICE (p>0.999) or between
CWI and ICE (p = 0.263, Table 1). There was no condition effect for the mean HR response
during the exercise (p = 0.679, Table 1). CWI decreased whole body sweat loss relative to CON
(p = 0.012) and ICE (p = 0.011); however, there was no condition effect for LSRarm (p = 0.994)
or LSRth (p = 0.592).
Tsk, Tre, Tre-Tsk gradient and thermal sensation. Significant interaction effect was
observed for Tre (p = 0.003) such that it was lower by ~0.3˚C in ICE versus CWI and CON dur-
ing the first 5–10 min of exercise (Fig 3A). However, the magnitude of increase in Tre during
the 60 min of exercise was greatest in ICE compared with CON (p = 0.003) and CWI
(p<0.001, Table 1). An interaction effect was observed for Tsk (p<0.001). CWI decreased Tsk
for up to 15–20 min during the exercise compared with CON and ICE (p<0.05, Fig 3B), while
ICE increased Tsk during the first 5 min of exercise compared with CON (p<0.05). Tre-Tsk gra-
dient depicted an interaction effect (p<0.001) whereby it increased during the first 15–20 min
of exercise in CWI relative to CON and ICE (p<0.05, Fig 3C). ICE decreased Tre-Tsk gradient
during the first 5 min of exercise compared with CON (p<0.05). An interaction effect was
observed for thermal sensation (p<0.001). Both CWI and ICE reduced thermal sensation dur-
ing the exercise compared with CON (p<0.05), but final values were not different between
conditions (p>0.05, Fig 3D). CWI decreased thermal sensation during the first 15 min of exer-
cise and between 30 to 35 min during the exercise, when compared with ICE (p<0.05).
NIRS data and skin PU. An interaction effect was evident for tHb (p<0.001) in which it
decreased during CWI and for up to 10–15 min during the exercise compared with CON and
ICE (p<0.05, Fig 4A). Significant interaction effect was observed for OxyHb (p = 0.005)
whereby it decreased during CWI for up to 15–20 min during the exercise compared with
CON and ICE (p<0.05, Fig 4B). OxyHb was also lower at the end of cooling during ICE com-
pared with CON (p = 0.034). Hb increased during the exercise (p<0.001) but there was no
Fig 1. Tre (A), Tsk (B), and Tre-Tsk gradient (C) during 60 min of cycling at fixed intensity (study 1). CON, control;
CWI, cold water immersion; ICE, ice slushy ingestion; � p<0.05 CWI versus CON; �� p<0.05 ICE versus CON; †
p<0.05 CWI versus ICE. Data are mean ± SD for n = 11 for Tsk. Due to missing data at certain time points during the
exercise, data for Tre and Tre-Tsk gradient are n = 10 during the first 10 min of exercise for all conditions and n = 8 or 9
thereafter (see S1 and S2 Tables for clarification).
https://doi.org/10.1371/journal.pone.0212966.g001
ICE and CWI on thermoregulatory behavior
PLOS ONE | https://doi.org/10.1371/journal.pone.0212966 February 27, 2019 8 / 19
Page 9
ICE and CWI on thermoregulatory behavior
PLOS ONE | https://doi.org/10.1371/journal.pone.0212966 February 27, 2019 9 / 19
Page 10
main condition (p>0.999) or interaction effect (p = 0.546, Fig 4C). No effect for condition
(p>0.05) or interaction (p>0.05) was observed for TOI, although it decreased during the exer-
cise (p<0.001, Fig 4D).
Skin PU depicted an interaction effect (p = 0.005) whereby both CWI (p = 0.012) and ICE
(p = 0.044) decreased PU by the end of cooling relative to CON (Fig 4E). An interaction effect
was evident for CVC (p = 0.001). CVC was not different between conditions at baseline
(p>0.05, CON: 0.39 ± 0.17 PU�mmHg-1, CWI: 0.49 ± 0.40 PU�mmHg-1, ICE: 0.37 ± 0.23
PU�mmHg-1). At the end of cooling, CWI (p<0.001) and ICE (p = 0.001) decreased CVC rela-
tive to CON (CON: 0.54 ± 0.20 PU�mmHg-1, CWI: 0.19 ± 0.11 PU�mmHg-1, ICE: 0.22 ± 0.10
PU�mmHg-1). When expressed as percentage of the baseline values, the changes in CVC were
+15%, -61% and -40% for CON, CWI and ICE, respectively.
Sweat threshold and sweat sensitivity from studies 1 and 2
CWI (p<0.001) and ICE (p = 0.019) delayed sweat recruitment in terms of exercise time com-
pared with CON (Table 2). Additionally, CWI delayed sweating onset by ~4 min relative to
ICE (p<0.001). Sweating occurred at a lower Tre in ICE versus CON (p = 0.007) and CWI
(p<0.001), while CWI resulted in an elevated Tre threshold for sweating compared with CON
(p = 0.007, Table 2). However, there was no difference between conditions for the Tb threshold
(p = 0.973). Tre sweat sensitivity was increased following CWI compared with CON (p =
0.007) and ICE (p = 0.009). There was a significant effect for Tb sweat sensitivity (Table 2).
CWI resulted in a higher sweat sensitivity compared with CON (p = 0.037), while the differ-
ence between CON and ICE did not reach statistical significance (p = 0.107).
Discussion
The present study compared the effects of precooling internally via ICE and externally via CWI
on thermoregulatory responses (e.g., Tsk and Tre) during steady state exercise (study 1) and
thermoregulatory behavior (i.e., power output and total work output) during the RPE clamp
protocol (study 2). These cooling methods were utilised for their distinct influences on Tsk and
Tre. As hypothesised, study 1 demonstrated that ICE resulted in a narrower Tre-Tsk gradient via
a reduction in Tre without influencing Tsk, and CWI increased the Tre-Tsk gradient via reduc-
tion in Tsk. Additionally, CWI significantly reduced whole body sweat loss and muscle blood
volume (i.e., tHb) but did not impair O2 utilisation as inferred from the TOI and Hb responses.
The main findings from study 2 showed that CWI increased MPO and total work output during
60 min of cycling regulated at RPE 15 when compared with CON only. Additionally, both pre-
cooling strategies improved thermal sensation compared with CON, with a longer and larger
effect observed following CWI. These changes occurred at similar HR response, muscle O2 utili-
sation (TOI and Hb) and skin PU between conditions. The LSR data from both studies showed
that CWI and ICE delayed sweating by 1–5 min relative to CON. Furthermore, CWI resulted in
a higher Tre threshold for sweating whereas ICE resulted in a lower Tre threshold for the effector
response; however, sweating occurred at similar Tb between conditions.
ICE decreased Tre by ~0.3˚C while CWI significantly decreased Tsk compared with CON
and ICE in both studies (Figs 1 and 3). In study 2, the higher work output observed during the
Fig 2. Changes in tHb (A), HbO2 (B), Hb (C) and TOI (D) during 60 min of cycling at fixed intensity (study 1).
CON, control; CWI, cold water immersion; ICE, ice slushy ingestion; � p<0.05 CWI versus CON; �� p<0.05 ICE
versus CON; † p<0.05 CWI versus ICE. Data are expressed as absolute changes from the baseline values and are
mean ± SD for n = 11, except for the final 20 min of exercise during CON and the final 10 min of exercise during ICE
where n = 10 due to probe damage.
https://doi.org/10.1371/journal.pone.0212966.g002
ICE and CWI on thermoregulatory behavior
PLOS ONE | https://doi.org/10.1371/journal.pone.0212966 February 27, 2019 10 / 19
Page 11
ICE and CWI on thermoregulatory behavior
PLOS ONE | https://doi.org/10.1371/journal.pone.0212966 February 27, 2019 11 / 19
Page 12
CWI trials was in agreement with the importance of Tsk as controller of thermoregulatory
behavior during exercise in the heat [4, 29]. Notably, manipulating Tsk via different ambient
temperatures (18˚C, 26˚C, 34˚C, and 42˚C) did not influence Tre but resulted in impaired
exercise capacity at a higher Tsk [29]. Similarly, self-selected work rate during 60 min of cycling
has also been shown to parallel the changes in Tsk rather than Tre [4]. Indeed, Tsk appeared to
have a greater influence on thermoregulatory behavior than Tre since the reduction in Tre via
ICE did not influence total work output when compared with CON in study 2. Nevertheless,
this current observation should be viewed with the understanding that Tre as an index of inner
body temperature is confounded by its delayed response.
Tsk and thermal perception have been identified as independent controllers for thermoreg-
ulatory behavior [1]. However, study 2 showed that decreased thermal sensation in ICE did
not have any beneficial effect on the work capacity when RPE was clamped at 15. Our findings
also contradicted previous works which demonstrated that L-Menthol mouth rinse and face
cooling improved thermal perception and increased time to exhaustion during a RPE clamp
exercise without concomitant changes in the physical thermal state (i.e., Tsk and Tre) [3, 5]. It
should be noted that maximum exercise duration within these two studies was less than 30
min, which paralleled the time course of changes in thermal sensation between ICE and CON
in study 2 (Fig 3D). Taken together, the current findings do not refute the importance of ther-
mal perception as controller of thermoregulatory behavior but suggest that the psychophysio-
logical effect of ICE and other cooling methods are most important during shorter trials. A
larger heat storage capacity conferred by ICE as a precooling strategy is negated by a lower
evaporative heat loss [30], which helps to explain the lack of a clear ergogenic effect of ICE on
endurance performance herein and within the literature [6]. In contrast, CWI significantly
improved thermal sensation when compared with CON and ICE for up to 35–40 min during
the RPE clamp exercise (study 2), supporting the importance of Tsk and thermal perception
for mediating thermoregulatory behavior.
In line with our initial hypothesis, ICE decreased the Tre-Tsk gradient whereas CWI signifi-
cantly increased the Tre-Tsk gradient (Figs 1 and 3). A high skin temperature (>35˚C) and a
concerted decrease in the Tre-Tsk gradient increase skin blood flow demand for heat dissipa-
tion [31]. CWI may alleviate cardiovascular strain by reducing reliance on cutaneous vasodila-
tion for heat loss [7]. Indeed, study 1 showed that CWI resulted in a marginal decrease of ~5
bpm in mean HR compared with CON during the exercise, consistent with others who
observed a transient decrease in the HR response to exercise following CWI at 17.7˚C x 30
min [32]. However, a narrower Tre-Tsk gradient during early exercise stages following ICE did
not affect the HR response compared with CON, which may be due to a greater increase in
Tre, causing the Tre-Tsk gradient in the ICE trials to increase rapidly to levels similar to the
CON trials. Another interesting observation was that although ICE and CWI decreased skin
PU before the exercise, there were no differences between all conditions during the RPE clamp
exercise in study 2 (Fig 4). For the ICE trials, this was concurrent with a higher Tsk during the
first 5 min of exercise compared with CON (Fig 3), which may be related to the narrow tem-
perature gradient between the environment, core body and skin [33], and attenuated evapora-
tive heat loss secondary to a delayed sweating onset (Table 2). The cooling effect of CWI on
the Tsk and Tre-Tsk gradient was observed for up to 15–20 min during the exercise (Fig 3),
Fig 3. Tre (A), Tsk (B), and Tre-Tsk gradient (C), and thermal sensation (D) during 60 min of cycling at RPE 15
(study 2). CON, control; CWI, cold -water immersion; ICE, ice slushy ingestion; � p<0.05 CWI versus CON; ��
p<0.05 ICE versus CON; † p<0.05 CWI versus ICE. Data are mean ± SD for n = 11 unless otherwise stated. Due to
missing data at certain time points during the exercise, data for Tre and Tre-Tsk gradient are n = 11 during the first 25
min of exercise for all conditions, and n = 10 for CWI and ICE thereafter (see S3 and S4 Tables for clarification).
https://doi.org/10.1371/journal.pone.0212966.g003
ICE and CWI on thermoregulatory behavior
PLOS ONE | https://doi.org/10.1371/journal.pone.0212966 February 27, 2019 12 / 19
Page 13
ICE and CWI on thermoregulatory behavior
PLOS ONE | https://doi.org/10.1371/journal.pone.0212966 February 27, 2019 13 / 19
Page 14
thereby maintaining the convective heat flux from the core body to the skin during early exer-
cise phases. However, as the environmental temperature was much higher than Tsk following
CWI, convective heat flux to the environment was most likely reversed until Tsk approximated
the environmental temperature, which may help to explain the lack of differences between
conditions in skin PU during the exercise.
ICE purportedly improves gross efficiency and has a glycogen sparing effect related to the
core body cooling effect [34]. However, no differences in the NIRS parameters were observed
in both studies (Figs 2 and 4), which suggested that ICE did not affect muscle metabolism.
Conversely, CWI decreased muscle blood volume and limited O2 delivery to locomotive mus-
cle during the precooling period and early exercise stages in both studies. It has been previ-
ously demonstrated that 5–15 min of CWI at 10˚C impaired muscle blood volume and local
tissue oxygenation [9, 10], resulting in greater anaerobic contribution during intermittent
high-intensity exercise in temperate environments [10]. Contrary to these earlier studies, the
present study provided evidence that CWI-induced decrease in muscle blood volume did not
adversely affect O2 utilisation during the steady state exercise (study 1) or the RPE clamp exer-
cise (study 2) as inferred from the changes in the TOI and Hb, attributable to a more thermally
comfortable CWI at 22˚C versus CWI at 10˚C. Importantly, a higher work output was
achieved during the CWI trials at muscle O2 utilisation and HR response similar to CON, con-
sistent with the propositions that relative exercise intensity was maintained secondary to allevi-
ated cardiovascular strain [35], and that individuals paced themselves such that metabolic
homeostasis was maintained [22]. It is also possible that a certain magnitude of the increase in
muscle blood volume during exercise in the no-cooling trials is related to elevated heat stress
and not to metabolic demand per se. In support, it has been shown that heat stress at rest and
during exercise increases blood flow to the muscle vasculature related to a direct thermal
response, an increase in arterial plasma adenosine triphosphate and/or modulation by the
muscle sympathetic vasoconstrictor activity [36]. Regardless, the present results are delimited
to submaximal exercise, and do not refute the possible effects that CWI may have on muscle
metabolic inertia during high-intensity exercise [10].
Both CWI and ICE delayed sweating response in terms of exercise time, in agreement with
previous precooling studies [32, 37, 38]. We found that sweat recruitment occurred at a lower Tre
in ICE and at a higher Tre in CWI compared with CON (Table 2), whereas others observed
Fig 4. Changes in tHb (A), OxyHb (B), Hb (C) TOI (D), and skin PU (E) during 60 min of cycling at RPE 15
(study 2). CON, control; CWI, cold water immersion; ICE, ice slushy ingestion; � p<0.05 CWI versus CON; �� p<0.05
ICE versus CON; † p<0.05 CWI versus ICE. Data are expressed as absolute changes from the baseline values and are
mean ± SD for n = 11, except for the final 30 min of exercise during CWI where n = 10 due to probe damage.
https://doi.org/10.1371/journal.pone.0212966.g004
Table 2. Tre and Tb at the onset of sweating and slopes of regression lines determined after plotting average LSR against Tre and Tb during exercise at fixed intensity
(study 1) and during the RPE clamp exercise (study 2).
CON CWI ICE P-value
Onset of sweating (min) 1.8 ± 1.8� # 6.4 ± 2.2# 2.7 ± 1.6� <0.001
Tre sweat threshold (˚C) 37.0 ± 0.3� # 37.1 ± 0.2# 36.8 ± 0.3� <0.001
Tb sweat threshold (˚C) 36.5 ± 0.3 36.5 ± 0.3 36.5 ± 0.4 0.973
Tre sweat sensitivity (mg�cm-2�min-1�˚C-1) 1.51 ± 0.61� 1.89 ± 0.82 1.42 ± 0.68� 0.001
Tb sweat sensitivity (mg�cm-2�min-1�˚C-1) 1.18 ± 0.42� 1.36 ± 0.55 1.39 ± 0.63 0.007
CON, control; CWI, cold water immersion; ICE, ice slushy ingestion; Tre, rectal temperature; Tb, weighted mean body; LSR, average local sweat rate for arm and thigh
� p<0.05 versus CWI# p<0.05 versus ICE. Data are mean ± SD for n = 19 from study 1 and 2. See S5 Table for effect sizes (Cohen’s d) calculated from mean differences between conditions
and pooled SD.
https://doi.org/10.1371/journal.pone.0212966.t002
ICE and CWI on thermoregulatory behavior
PLOS ONE | https://doi.org/10.1371/journal.pone.0212966 February 27, 2019 14 / 19
Page 15
unchanged Tre sweating threshold [32], or lower Tb sweating threshold determined from weighted
Tsk and esophageal temperature [39] after cooling. While the disparity may be due to the different
core temperature indexes or different cooling techniques (cold air versus CWI), it is more likely
attributable to the influence that skin and core temperatures have on the sweat response. Indeed,
sweating occurred at similar Tb for all conditions and the differences in sweat gain between condi-
tions were attenuated when plotted against Tb (Table 2). Because ICE primarily activates the sudo-
motor response via the abdominal thermoreceptors [16], it is not surprising that the sweat gain
against Tb and Tre are similar, and the difference in sweat sensitivity between CON and ICE did
not reach statistical significance. Conversely, CWI modifies the sweat responses via the cutaneous
afferent signals and the nitric oxide pathway [14]. Because Tre was similar between CWI and
CON in studies 1 and 2, we suggest that the higher sweat gain following CWI reflects the rapid
increase in Tsk during the exercise. Taken together, our data demonstrate that precooling attenu-
ates the thermal afferent signals such that the efferent signals start firing when a specific Tb thresh-
old is attained, but CWI modifies the efferent signals by the changes in Tsk.
Limitations
The NIRS parameters derived using the modified Beer-Lambert method are known to be affected
by skin blood flow, whereas TOI is based on the spatially-resolved spectroscopy and has been
shown to be affected by a lesser degree [28, 40]. It is important to note that the aforementioned
studies examined the NIRS signals during heat stress at rest or a brief bout of single joint exercise.
In contrast, skin blood flow has been shown to have a lesser influence on the NIRS signal during
CWI [41]. If the cutaneous interference did contribute to the NIRS signal, it seems logical to have
both parameters change similarly by cooling or heating. Moreover, skin and muscle blood flow
increase drastically during prolonged exercise in the heat, and muscle blood flow has been shown
to increase during local heating [42]. Therefore, while the contribution of skin blood flow to the
NIRS signals during exercise cannot be dismissed, it remains challenging to distinguish between
thermoregulatory and metabolic demands from the cutaneous interference.
CVC is often expressed as percentage of maximum cutaneous vasodilation achieved by
local heating and/or administration of sodium nitroprusside. However, the reliability of base-
line CVC, expressed as absolute values or relative to maximum CVC, remains poor with coeffi-
cients of variation ranged from 25–30% [43]. Improved reliability has been observed as CVC
increases during local heating [43] or during exercise [44]. Moreover, our pilot observation
showed that it took more than 40 min to achieve a plateau in CVC after the exercise. As such,
skin PU data are expressed in absolute values.
The difference in power output between CON and CWI in study 2 is considered as small
effect (Cohen’s d = 0.42, mean difference of 8 W). The modest beneficial effect from CWI may
be related to the training status of the present cohort ( _VO2peak <55 mL.kg-1.min-1), as well-
trained athletes have been shown to have a greater beneficial effect from precooling maneuvers
[8]. Additionally, the present results are based on a male sample. Future research should
explore the possible gender effect on precooling and the resultant physiological responses, as it
has been shown that females have a lower evaporative heat loss and blunted sweating thermo-
sensitivity than males during exercise at a fixed rate of metabolic heat production after control-
ling for menstrual cycle, body mass, and _VO2peak relative to lean muscle mass [45].
Conclusions
To conclude, significant skin cooing by CWI resulted in a greater reduction in thermal sensa-
tion and improved thermoregulatory behavior (i.e., higher MPO and total work output).
ICE and CWI on thermoregulatory behavior
PLOS ONE | https://doi.org/10.1371/journal.pone.0212966 February 27, 2019 15 / 19
Page 16
Additionally, although CWI decreased muscle blood volume during early exercise stages, it
did not limit muscle oxygenation during steady state exercise or RPE clamp exercise in the
heat. Conversely, a reduction of ~0.3˚C in Tre by ICE decreased thermal sensation but did not
confer any ergogenic effect on the thermoregulatory behavior during 60 min of exercise. As
such, the duration of events should be taken into consideration when planning for cooling
strategies prior to exercise in hot conditions. ICE and CWI delayed sweat recruitment in terms
of exercise time and attenuated the thermal efferent signals until a specific Tb threshold was
attained, but the efferent signals were modified by the changes in Tsk in CWI.
Supporting information
S1 Table. Rectal temperature (Tre) response during 60 min of cycling at fixed intensity fol-
lowing 30 min of precooling (study 1). CON, control; CWI, cold water immersion, ICE, ice
slushy ingestion.
(PDF)
S2 Table. Rectal-to-skin temperature (Tre-Tsk) gradient response during 60 min of cycling
at fixed intensity following 30 min of precooling (study 1). CON, control; CWI, cold water
immersion, ICE, ice slushy ingestion.
(PDF)
S3 Table. Rectal temperature (Tre) response during 60 min of cycling at RPE 15 following
30 min of precooling (study 2). CON, control; CWI, cold water immersion, ICE, ice slushy
ingestion.
(PDF)
S4 Table. Rectal-to-skin temperature (Tre-Tsk) gradient response during 60 min of cycling
at RPE 15 following 30 min of precooling (study 2). CON, control; CWI, cold water immer-
sion, ICE, ice slushy ingestion.
(PDF)
S5 Table. Effect sizes (Cohen’s d) calculated for the mean physiological and perceptual
responses, rectal temperature (Tre) and weighted mean body temperature (Tb) at the onset
of sweating and sweat sensitivity during steady state exercise (study 1) and during the RPE
clamp exercise (study 2). CON, control; CWI, cold water immersion, ICE, ice slushy inges-
tion.
(PDF)
S1 Dataset. Data underlying the findings reported in the present manuscript. CON, con-
trol; CWI, cold water immersion, ICE, ice slushy ingestion.
(XLSX)
Acknowledgments
The authors would like to express our gratitude to the participants who took part in the study.
Author Contributions
Conceptualization: Hui C. Choo, Jeremiah J. Peiffer, Chris R. Abbiss.
Data curation: Hui C. Choo, João P. Lopes-Silva, Ricardo N. O. Mesquita.
Formal analysis: Hui C. Choo, João P. Lopes-Silva, Chris R. Abbiss.
Methodology: Hui C. Choo, Tatsuro Amano, Narihiko Kondo.
ICE and CWI on thermoregulatory behavior
PLOS ONE | https://doi.org/10.1371/journal.pone.0212966 February 27, 2019 16 / 19
Page 17
Supervision: Jeremiah J. Peiffer, Chris R. Abbiss.
Writing – original draft: Hui C. Choo.
Writing – review & editing: Hui C. Choo, Jeremiah J. Peiffer, João P. Lopes-Silva, Tatsuro
Amano, Narihiko Kondo, Chris R. Abbiss.
References1. Schlader ZJ, Stannard SR, Mundel T. Human thermoregulatory behavior during rest and exercise—A
prospective review. Physiol Behav. 2010; 99(3):269–75. https://doi.org/10.1016/j.physbeh.2009.12.003
PMID: 20006632
2. Flouris AD, Cheung SS. Human conscious response to thermal input is adjusted to changes in mean
body temperature. Br J Sports Med. 2008; 43(3):199–203. https://doi.org/10.1136/bjsm.2007.044552
PMID: 18216157
3. Schlader ZJ, Simmons SE, Stannard SR, Mundel T. The independent roles of temperature and thermal
perception in the control of human thermoregulatory behavior. Physiol Behav. 2011; 103(2):217–24.
WOS:000290011300013. https://doi.org/10.1016/j.physbeh.2011.02.002 PMID: 21315099
4. Schlader ZJ, Simmons SE, Stannard SR, Mundel T. Skin temperature as a thermal controller of exer-
cise intensity. Eur J Appl Physiol. 2011; 111(8):1631–9. WOS:000293980000009. https://doi.org/10.
1007/s00421-010-1791-1 PMID: 21197543
5. Flood TR, Waldron M, Jeffries O. Oral L-menthol reduces thermal sensation, increases work-rate and
extends time to exhaustion, in the heat at a fixed rating of perceived exertion. Eur J Appl Physiol. 2017;
117(7):1501–12. https://doi.org/10.1007/s00421-017-3645-6 PMID: 28508114
6. Choo HC, Nosaka K, Peiffer JJ, Ihsan M, Abbiss CR. Ergogenic effects of precooling with cold water
immersion and ice ingestion: A meta-analysis. Eur J Sport Sci. 2017; 18(2):170–81. https://doi.org/10.
1080/17461391.2017.1405077 PMID: 29173092
7. Bongers CC, Hopman MT, Eijsvogels TM. Cooling interventions for athletes: An overview of effective-
ness, physiological mechanisms, and practical considerations. Temperature. 2017; 4(1):60–78. https://
doi.org/10.1080/23328940.2016.1277003
8. Wegmann M, Faude O, Poppendieck W, Hecksteden A, Frohlich M, Meyer T. Pre-cooling and sports
performance: A meta-analytical review. Sports Med. 2012; 42(7):545–64. Epub 2012/05/31. https://doi.
org/10.2165/11630550-000000000-00000 PMID: 22642829.
9. Ihsan M, Watson G, Lipski M, Abbiss CR. Influence of postexercise cooling on muscle oxygenation and
blood volume changes. Med Sci Sports Exerc. 2013; 45(5):876–82. https://doi.org/10.1249/MSS.
0b013e31827e13a2 PMID: 23247707
10. Stanley J, Peake JM, Coombes JS, Buchheit M. Central and peripheral adjustments during high-inten-
sity exercise following cold water immersion. Eur J Appl Physiol. 2014; 114(1):147–63. https://doi.org/
10.1007/s00421-013-2755-z PMID: 24158407
11. Periard JD, Thompson MW, Caillaud C, Quaresima V. Influence of heat stress and exercise intensity on
vastus lateralis muscle and prefrontal cortex oxygenation. Eur J Appl Physiol. 2013; 113(1):211–22.
https://doi.org/10.1007/s00421-012-2427-4 PMID: 22648526
12. Proulx CI, Ducharme MB, Kenny GP. Safe cooling limits from exercise-induced hyperthermia. Eur J
Appl Physiol. 2006; 96(4):434–45. https://doi.org/10.1007/s00421-005-0063-y PMID: 16341523
13. Siegel R, Laursen PB. Keeping your cool: Possible mechanisms for enhanced exercise performance in
the heat with internal cooling methods. Sports Med. 2012; 42(2):89–98. https://doi.org/10.2165/
11596870-000000000-00000 PMID: 22175533
14. Fujii N, McGinn R, Halili L, Singh MS, Kondo N, Kenny GP. Cutaneous vascular and sweating
responses to intradermal administration of ATP: A role for nitric oxide synthase and cyclooxygenase? J
Physiol. 2015; 593(11):2515–25. https://doi.org/10.1113/JP270147 PMID: 25809194
15. Shibasaki M, Crandall CG. Mechanisms and controllers of eccrine sweating in humans. Front Biosci
(Scholar Ed). 2010; 2:685–96.
16. Morris NB, Bain AR, Cramer MN, Jay O. Evidence that transient changes in sudomotor output with cold
and warm fluid ingestion are independently modulated by abdominal, but not oral thermoreceptors. J
Appl Physiol. 2014; 116(8):1088–95. https://doi.org/10.1152/japplphysiol.01059.2013 PMID: 24577060
17. Morris NB, Coombs G, Jay O. Ice slurry ingestion leads to a lower net heat loss during exercise in the
heat. Med Sci Sports Exerc. 2016; 48(1):114–22. https://doi.org/10.1249/MSS.0000000000000746
PMID: 26258857
ICE and CWI on thermoregulatory behavior
PLOS ONE | https://doi.org/10.1371/journal.pone.0212966 February 27, 2019 17 / 19
Page 18
18. Tucker R, Marle T, Lambert EV, Noakes TD. The rate of heat storage mediates an anticipatory reduc-
tion in exercise intensity during cycling at a fixed rating of perceived exertion. J Physiol. 2006; 574
(3):905–15. https://doi.org/10.1113/jphysiol.2005.101733
19. Siegel R, Mate J, Watson G, Nosaka K, Laursen PB. Pre-cooling with ice slurry ingestion leads to simi-
lar run times to exhaustion in the heat as cold water immersion. J Sports Sci. 2012; 30(2):155–65.
https://doi.org/10.1080/02640414.2011.625968 PMID: 22132792
20. Kuipers H, Verstappen F, Keizer H, Geurten P, Van Kranenburg G. Variability of aerobic performance in
the laboratory and its physiologic correlates. Int J Sports Med. 1985; 6(04):197–201. https://doi.org/doi.
org/10.1055/s-2008-1025839
21. Borg GA. Psychophysical bases of perceived exertion. Med Sci Sports Exerc. 1982; 14(5):377–81.
PMID: 7154893
22. Lander PJ, Butterly RJ, Edwards AM. Self-paced exercise is less physically challenging than enforced
constant pace exercise of the same intensity: Influence of complex central metabolic control. Br J Sports
Med. 2009; 43(10):789–95. https://doi.org/10.1136/bjsm.2008.056085 PMID: 19196729
23. Young AJ, Sawka MN, Epstein Y, DeCristofano B, Pandolf KB. Cooling different body surfaces during
upper and lower body exercise. J Appl Physiol. 1987; 63(3):1218–23. https://doi.org/10.1152/jappl.
1987.63.3.1218 PMID: 3654466
24. Ramanathan NL. A new weighting system for mean surface temperature of the human body. J Appl
Physiol. 1964; 19(3):531–3.
25. Colin J, Timbal J, Houdas Y, Boutelier C, Guieu JD. Computation of mean body temperature from rectal
and skin temperatures. J Appl Physiol. 1971; 31(3):484–9. https://doi.org/10.1152/jappl.1971.31.3.484
PMID: 5111868
26. Kenefick RW, Cheuvront SN, Elliott LD, Ely BR, Sawka MN. Biological and analytical variation of the
human sweating response: Implications for study design and analysis. Am J Physiol Regul Integr Comp
Physiol. 2012; 302(2):R252–R8. https://doi.org/10.1152/ajpregu.00456.2011 PMID: 22071159
27. Cheuvront SN, Bearden SE, Kenefick RW, Ely BR, DeGroot DW, Sawka MN, et al. A simple and valid
method to determine thermoregulatory sweating threshold and sensitivity. J Appl Physiol. 2009; 107
(1):69–75. https://doi.org/10.1152/japplphysiol.00250.2009 PMID: 19423839
28. Messere A, Roatta S. Influence of cutaneous and muscular circulation on spatially resolved versus stan-
dard Beer–Lambert near-infrared spectroscopy. Physiol Rep. 2013; 1(7):e00179. https://doi.org/10.
1002/phy2.179 PMID: 24744858
29. Cuddy JS, Hailes WS, Ruby BC. A reduced core to skin temperature gradient, not a critical core temper-
ature, affects aerobic capacity in the heat. J Therm Biol. 2014; 43:7–12. https://doi.org/10.1016/j.
jtherbio.2014.04.002 PMID: 24956952
30. Jay O, Morris NB. Does cold water or ice slurry ingestion during exercise elicit a net body cooling effect
in the heat? Sports Med. 2018:1–13.
31. Sawka MN, Cheuvront SN, Kenefick RW. High skin temperature and hypohydration impair aerobic per-
formance. Exp Physiol. 2012; 97(3):327–32. https://doi.org/10.1113/expphysiol.2011.061002 PMID:
22143882
32. Wilson TE, Johnson SC, Petajan JH, Davis SL, Gappmaier E, Luetkemeier MJ, et al. Thermal regula-
tory responses to submaximal cycling following lower-body cooling in humans. Eur J Appl Physiol.
2002; 88(1–2):67–75. https://doi.org/10.1007/s00421-002-0696-z PMID: 12436272
33. Taylor NA. Challenges to temperature regulation when working in hot environments. Ind Health. 2006;
44(3):331–44. PMID: 16922177
34. Zimmermann MR, Landers GJ, Wallman KE, Saldaris J. The effects of crushed ice ingestion prior to
steady state exercise in the heat. Int J Sport Nutr Exerc Metab. 2017:1–21. https://doi.org/10.1123/
ijsnem.2017-0057
35. Periard JD, Cramer MN, Chapman PG, Caillaud C, Thompson MW. Cardiovascular strain impairs pro-
longed self-paced exercise in the heat. Exp Physiol. 2011; 96(2):134–44. https://doi.org/10.1113/
expphysiol.2010.054213 PMID: 20851861
36. Pearson J, Low DA, Stohr E, Kalsi K, Ali L, Barker H, et al. Hemodynamic responses to heat stress in
the resting and exercising human leg: Insight into the effect of temperature on skeletal muscle blood
flow. Am J Physiol Regul Integr Comp Physiol. 2011; 300(3):R663–R73. https://doi.org/10.1152/
ajpregu.00662.2010 PMID: 21178127
37. Zimmermann MR, Landers GJ, Wallman KE. Crushed ice ingestion does not improve female cycling
time trial performance in the heat. Int J Sport Nutr Exerc Metab. 2017:1–23. Epub 2016/07/28. https://
doi.org/10.1123/ijsnem.2016-0028 PMID: 27459723.
38. Morrison SA, Cheung S, Cotter JD. Importance of airflow for physiologic and ergogenic effects of pre-
cooling. J Athl Train. 2014; 49(5):632–9. https://doi.org/10.4085/1062-6050-49.3.27 PMID: 25144598
ICE and CWI on thermoregulatory behavior
PLOS ONE | https://doi.org/10.1371/journal.pone.0212966 February 27, 2019 18 / 19
Page 19
39. Olschewski H, Bruck K. Thermoregulatory, cardiovascular, and muscular factors related to exercise
after precooling. J Appl Physiol. 1988; 64(2):803–11. https://doi.org/10.1152/jappl.1988.64.2.803
PMID: 3372438
40. Tew GA, Ruddock AD, Saxton JM. Skin blood flow differentially affects near-infrared spectroscopy-
derived measures of muscle oxygen saturation and blood volume at rest and during dynamic leg exer-
cise. Eur J Appl Physiol. 2010; 110(5):1083–9. https://doi.org/10.1007/s00421-010-1596-2 PMID:
20700602
41. Choo HC, Nosaka K, Peiffer JJ, Ihsan M, Yeo CC, Abbiss CR. Peripheral blood flow changes in
response to post-exercise cold water immersion. Clin Physiol Funct Imaging. 2016; 38(1):46–55.
https://doi.org/10.1111/cpf.12380 PMID: 27464622
42. Heinonen I, Brothers RM, Kemppainen J, Knuuti J, Kalliokoski KK, Crandall CG. Local heating, but not
indirect whole body heating, increases human skeletal muscle blood flow. J Appl Physiol. 2011; 111
(3):818–24. https://doi.org/10.1152/japplphysiol.00269.2011 PMID: 21680875
43. Tew GA, Klonizakis M, Moss J, Ruddock AD, Saxton JM, Hodges GJ. Reproducibility of cutaneous ther-
mal hyperaemia assessed by laser Doppler flowmetry in young and older adults. Microvasc Res. 2011;
81(2):177–82. https://doi.org/10.1016/j.mvr.2010.12.001 PMID: 21167843
44. Choo HC, Nosaka K, Peiffer JJ, Ihsan M, Yeo CC, Abbiss CR. Reliability of laser Doppler, near-infrared
spectroscopy and Doppler ultrasound for peripheral blood flow measurements during and after exercise
in the heat. J Sports Sci. 2016; 35(17):1715–23. https://doi.org/10.1080/02640414.2016.1235790
PMID: 27649579
45. Gagnon D, Kenny GP. Sex modulates whole-body sudomotor thermosensitivity during exercise. J Phy-
siol. 2011; 589(24):6205–17. https://doi.org/10.1113/jphysiol.2011.219220
ICE and CWI on thermoregulatory behavior
PLOS ONE | https://doi.org/10.1371/journal.pone.0212966 February 27, 2019 19 / 19